Corona Discharge

How corona discharge works

Hold a high-voltage conductor in air and the electric field is strongest right at its surface, falling off with distance. Air is normally a good insulator, but every cubic centimetre of it contains a few free electrons knocked loose by cosmic rays and natural radioactivity. When the surface field climbs past the breakdown gradient of air — about 30 kV/cm peak at sea level — those stray electrons get accelerated hard enough between collisions to knock additional electrons off neutral air molecules. Each freed electron does the same to the next, and you get an electron avalanche: a self-sustaining cascade of ionization confined to the thin, high-field sheath hugging the conductor.

That ionized sheath is corona. The recombination of ions and electrons, plus the relaxation of excited molecules, emits faint violet-blue light (mostly from excited nitrogen). The collisions heat the air slightly, producing pressure pulses you hear as a hiss and a 100/120 Hz hum. The discharge fires in short current pulses — Trichel pulses on the negative half-cycle, broader streamers on the positive — and those nanosecond current spikes radiate broadband radio interference. The ion chemistry breaks O₂ into ozone and oxidizes nitrogen into NOx, which is why a corona-active substation smells sharp and slightly chlorine-like.

Crucially, corona is self-limiting. Once the sheath ionizes, the cloud of ions it sheds drifts outward and partially screens the conductor, holding the discharge to a thin glow layer. The bulk of the air gap stays insulating. That's the whole difference between corona and a flashover arc: corona is a leak at the surface; an arc is a complete bridge across the gap.

The governing physics: Peek's law

The classic engineering model is Peek's law (F.W. Peek, 1911), an empirical fit for the visual corona-onset gradient on a smooth cylindrical conductor. The disruptive critical field at the conductor surface is:

Corona onset (visual) gradient on a smooth round conductor:

  E_v = E_0 · m · δ · (1 + K / sqrt(δ · r))      [kV/cm, peak]

where
  E_0 = 30 kV/cm        breakdown gradient of air at STP (≈ 21.1 kV/cm RMS)
  K   = 0.301           Peek's surface-curvature constant (cm^0.5)
  m   = surface factor  1.0 smooth/polished, ~0.8 to 0.87 stranded ACSR,
                        down to ~0.7 dirty/weathered, lower still when wet
  δ   = air density factor = (3.92 · b) / (273 + t)
        b = barometric pressure in cmHg, t = temperature °C
        δ = 1.0 at 25°C and 76 cmHg
  r   = conductor radius in cm

Two things fall straight out of this. First, the 1/sqrt(r) term means thin wires actually need a higher surface gradient to onset than fat ones — yet because a thin wire concentrates the field so strongly, it still coronas at a far lower voltage, so a sharp point is the worst case. Second, the δ term means thin air (high altitude, hot weather) lowers the onset, while dense cold air raises it. The line-to-neutral voltage at which corona starts, the critical disruptive voltage, is found by relating the surface field to the applied voltage through the line geometry:

Surface gradient from applied phase voltage V (single conductor,
spacing D, radius r, with D >> r):

  E_surface = V / ( r · ln(D / r) )          [kV/cm]

Critical disruptive voltage (onset):

  V_c = E_0 · m · δ · r · ln(D / r)          [kV to neutral, peak]

Peek's fair-weather corona power loss per phase per km:

  P = ( 241 / δ ) · (f + 25) · sqrt(r / D) · (V − V_c)^2 · 1e-5   [kW/km/phase]

  f = frequency (Hz),  V and V_c in kV to neutral (RMS)

The loss equation captures the headline behaviour: power dissipated rises with the square of how far the operating voltage exceeds the onset voltage. Push 10% over onset and loss is modest; push 40% over and it balloons. Designers therefore don't try to keep corona at exactly zero — they keep the working surface gradient comfortably below onset (typically 16 to 18 kV/cm RMS) so that (V − V_c) is negative or near-zero in fair weather, and only the rare foul-weather hours push it positive.

Worked example: a 400 kV line conductor

Take a 400 kV (≈ 231 kV to neutral) overhead line, single ACSR conductor, and check whether it coronas in fair weather at sea level:

Conductor radius:   r = 1.4 cm   ("Drake" ACSR, ~28 mm dia.)
Phase spacing:      D = 760 cm    (geometric mean ~7.6 m)
Surface factor:     m = 0.85      (stranded, fair weather)
Air density:        δ = 1.0       (sea level, 25°C)

Critical disruptive voltage (RMS to neutral):
  V_c = 21.1 · m · δ · r · ln(D/r)
      = 21.1 · 0.85 · 1.0 · 1.4 · ln(760/1.4)
      = 21.1 · 0.85 · 1.4 · ln(543)
      = 21.1 · 0.85 · 1.4 · 6.30
      ≈ 158 kV to neutral

Operating voltage to neutral:  231 kV  >  V_c (158 kV)  →  CORONA.

A single Drake conductor at 400 kV is over its corona-onset voltage — it would glow, hiss, and waste power continuously. That's exactly why no utility runs a single conductor per phase at 400 kV. Re-run it as a 2-conductor bundle spaced 450 mm apart and the phase's equivalent radius jumps to roughly r_eq = sqrt(r · s) = sqrt(1.4 · 45) ≈ 7.9 cm, dropping the surface gradient below onset and killing fair-weather corona. The fix isn't a bigger single wire — it's splitting the phase.

Design levers: how engineers suppress corona

• Increase effective radius. Corona-onset gradient is governed by surface field, and surface field falls as conductor radius rises. A bigger conductor coronas at higher voltage — but a solid 8 cm conductor is heavy, expensive, and uses far more aluminium than its current rating needs.
• Bundle the phase. The standard EHV answer. Two to four sub-conductors spaced ~300 to 500 mm apart make the phase behave like one large-diameter conductor at a fraction of the metal. 400 kV typically uses 2-bundle; 500 to 765 kV uses 3- or 4-bundle. Bundling also drops series reactance, raising transfer capacity.
• Corona / grading rings. Smooth toroids fitted at the high-field ends of insulator strings, bushings, and HV apparatus terminals. They enlarge the local radius and smooth equipotentials, pulling the field below onset where geometry would otherwise spike it.
• Smooth, clean surfaces. Burrs, nicks, scratches, dead insects, water droplets, and stranding all create local field-intensifying points (high m-factor loss). Hardware is deburred and rounded; conductor handling avoids scoring the surface.
• Grading caps & semiconductive coatings. Inside switchgear, cable terminations, and machine windings, stress-grading layers spread the field so no sharp edge reaches onset.

Real-world corona figures

System / situation	Conductor arrangement	Typical corona behaviour
132 kV line	Single conductor (~18 mm)	Little to no corona in fair weather; minor foul-weather loss
400 kV line	2-bundle (~28 mm subs)	Negligible fair-weather; few kW/km/phase in rain
500 kV line	3-bundle	Designed for low audible noise; rain loss tens of kW/km/phase
765 kV line	4- to 6-bundle	Audible-noise & RI-limited design; bundling dominates the spec
HVDC ±500 kV pole	Bundle + ion-current control	Corona produces space charge and ground-level ion currents
Sharp point / burr at a few kV	Needle electrode	Visible corona at only 2 to 10 kV — worst-case geometry
Spacecraft / aircraft HV	Exposed terminals at altitude	Onset drops with low pressure (Paschen) — strict spacing rules

To put numbers on the loss: a 500 kV, 400 km line might lose well under 1 MW total to corona averaged over a fair-weather year, but during a heavy rain event the instantaneous corona loss can spike past 5 to 10 MW across all three phases. Because the foul-weather hours are intense though infrequent, annual corona energy is specified statistically from local weather data, and the design target is usually a long-term average loss small relative to the line's I²R (resistive) loss — typically a few percent of total losses, not the dominant term, for a well-engineered EHV line.

When corona is the feature, not the bug

On power lines corona is purely a loss. But the same physics is deliberately exploited elsewhere:

• Electrostatic precipitators (ESPs). A negative-corona wire charges dust particles in a flue-gas stream so they migrate to grounded collector plates. ESPs clean the exhaust of nearly every coal plant and cement kiln, removing 99%+ of particulate by mass — corona is the charging engine.
• Laser printers and photocopiers. A corona wire (or charge roller) lays a uniform charge on the photoconductor drum before the laser writes the image; another corona transfers toner to paper.
• Ionizers and air purifiers. Corona generates ions that charge and agglomerate airborne particles.
• Ozone generators. Corona (or its cousin, dielectric-barrier discharge) is a standard industrial route to ozone for water and wastewater treatment.
• Corona surface treatment. Plastics like polyethylene film are passed under a corona to oxidize and roughen the surface so inks and adhesives bond — used in nearly all flexible-packaging printing.

The St. Elmo's fire seen on ship masts and aircraft wingtips in thunderstorms is natural corona: the storm's ambient field drives the sharp metal tips past onset.

Failure modes and why corona matters

• Continuous power loss. The headline cost on transmission lines — a permanent leak that scales with (V − V_c)². It never trips a protective relay; it just quietly burns energy, dominated by foul weather.
• Radio and TV interference (RI/TVI). The nanosecond current pulses radiate broadband noise across the AM and low-VHF bands. Lines near residential and broadcast areas are designed to RI limits set by standards such as CISPR and IEEE; a single defective fitting can generate complaints for kilometres.
• Audible noise. The hiss and 100/120 Hz hum are a community-nuisance and permitting issue, especially for 500 kV-and-above lines. Audible-noise limits often drive the bundle choice more than corona loss itself.
• Insulation degradation. Where corona occurs inside equipment — voids in cable insulation, gaps in machine windings, defects in bushings — the localized partial discharge plus ozone and NOx chemically attacks the dielectric. Over years this erodes the insulation and ends in breakdown. Partial-discharge (PD) testing per IEC 60270 hunts for exactly this.
• Ozone and NOx production. Around dense HV apparatus, corona-generated ozone corrodes rubber and some metals and is a respiratory hazard in poorly ventilated switchrooms.
• Precursor to flashover. Heavy corona signals an over-stressed region. If the field keeps rising, the glow can develop streamers that bridge the gap into a full arc — the corona was the early warning.

Common misconceptions and pitfalls

• "Corona starts at a fixed voltage." It starts at a fixed field gradient at the conductor surface (~30 kV/cm peak), not a fixed voltage. The voltage that produces that gradient depends entirely on conductor radius, surface finish, geometry, and air density. A needle coronas at a few kV; a fat bundle stays clear at 500 kV.
• "Corona is the same as an arc." No — corona is a thin, self-limiting partial discharge carrying microamps; an arc is full-gap breakdown carrying hundreds or thousands of amps in a hot plasma channel.
• "Bigger voltage is always the problem." Geometry usually dominates. A clean, well-radiused 765 kV bundle can corona less than a scratched, burred fitting on a 132 kV line. Surface defects and sharp edges are the real culprits.
• "Corona loss is constant." It's dominated by weather. Fair-weather loss is small; rain, fog, snow, and hoarfrost can raise it by one to two orders of magnitude. Any single-number "corona loss" without a weather basis is meaningless.
• "Corona rings are decorative." They're load-bearing field management. Remove the ring from a 230 kV insulator line-end and the local field spikes, restarting corona that degrades the insulator and radiates noise.
• "Higher altitude doesn't matter." It matters a lot — the air density factor δ falls with altitude, lowering onset voltage. Lines over high passes are corona-derated and often use heavier bundling than identical sea-level lines.